Social Media and the

NUMBERS, FACTS AND TRENDS SHAPING THE WORLD
AUGUST 26, 2014
Facebook, Twitter, and other platforms did not provide new outlets for the discussion
of the Snowden-NSA revelations. In fact, people were less likely to discuss these issues
on social media than they were in person and, if people thought their social media
friends and followers disagreed with them, they were less likely to want to discuss the
issues at all.
FOR FURTHER INFORMATION
ON THIS REPORT:
Keith N. Hampton, Associate Professor, Rutgers University
Lee Rainie, Director, Internet Project
Weixu Lu, PhD Student, Rutgers University
Maria Dwyer, PhD Student, Rutgers University
Inyoung Shin, PhD Student, Rutgers University
Kristen Purcell, Associate Director, Internet Project
202.419.4372
www.pewresearch.org
RECOMMENDED CITATION: Hampton, K.N., Rainie, L., Lu, W., Dwyer, M., Shin, I., & Purcell, K. (2014). “Social Media and the ‘Spiral of
Silence.’ Pew Research Center, Washington, DC.
Available at http://www.pewinternet.org/2014/08/26/social-media-and-the-spiral-of-silence/
1
PEW RESEARCH CENTER
About this Report
An informed citizenry depends on people’s exposure to information on important political issues
and on their willingness to discuss these issues with those around them. The rise of social media,
such as Facebook and Twitter, has introduced new spaces where political discussion and debate
can take place. This report explores the degree to which social media affects a long-established
human attribute—that those who think they hold minority opinions often self-censor, failing to
speak out for fear of ostracism or ridicule. It is called the “spiral of silence.”
This report is a collaborative effort based on the input and analysis of the following individuals:1
Keith N. Hampton, Associate Professor, Rutgers University
Lee Rainie, Director, Internet Project
Weixu Lu, PhD student, Rutgers University
Maria Dwyer, PhD student, Rutgers University
Inyoung Shin, PhD student, Rutgers University
Kristen Purcell, Associate Director for Research, Internet Project
Other major reports from the Pew Research Center Internet Project on the social and political
impact of social networking sites on social and political activity can be found at:
http://www.pewinternet.org/2012/10/19/social-media-and-political-engagement/
http://www.pewinternet.org/2012/09/04/politics-on-social-networking-sites/
http://www.pewinternet.org/2013/04/25/civic-engagement-in-the-digital-age/
http://www.pewinternet.org/2012/02/03/why-most-facebook-users-get-more-than-they-give/
http://www.pewinternet.org/2011/06/16/social-networking-sites-and-our-lives/
http://www.pewinternet.org/2009/11/04/social-isolation-and-new-technology/
http://www.pewinternet.org/2004/10/27/the-internet-and-democratic-debate/
We are grateful to the following individuals for their comments and advice as we developed this work: Pablo Boczkowski (Northwestrn
University), William Eveland (The Ohio State University), and Rima Wilkes (University of British Columbia).
1
www.pewresearch.org
2
PEW RESEARCH CENTER
About the Pew Research Center
Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes
and trends shaping America and the world. It does not take policy positions. It conducts public
opinion polling, demographic research, media content analysis and other empirical social science
research. The center studies U.S. politics and policy views; media and journalism; internet and
technology; religion and public life; Hispanic trends; global attitudes and U.S. social and demographic trends. All of the center’s reports are available at www.pewresearch.org. Pew Research
Center is a subsidiary of The Pew Charitable Trusts.
© Pew Research Center 2014
www.pewresearch.org
3
PEW RESEARCH CENTER
Summary of Findings
A major insight into human behavior from pre-internet era studies of communication is the
tendency of people not to speak up about policy issues in public—or among their family, friends,
and work colleagues—when they believe their own point of view is not widely shared. This
tendency is called the “spiral of silence.”2
Some social media creators and supporters have hoped that social media platforms like Facebook
and Twitter might produce different enough discussion venues that those with minority views
might feel freer to express their opinions, thus broadening public discourse and adding new
perspectives to everyday discussion of political issues.
We set out to study this by conducting a survey of 1,801 adults.3 It focused on one important public
issue: Edward Snowden’s 2013 revelations of widespread government surveillance of Americans’
phone and email records. We selected this issue because other surveys by the Pew Research Center
at the time we were fielding this poll showed that Americans were divided over whether the NSA
contractor’s leaks about surveillance were justified and whether the surveillance policy itself was a
good or bad idea. For instance, Pew Research found in one survey that 44% say the release of
classified information harms the public interest while 49% said it serves the public interest.
The survey reported in this report sought people’s opinions about the Snowden leaks, their
willingness to talk about the revelations in various in-person and online settings, and their
perceptions of the views of those around them in a variety of online and off-line contexts.
This survey’s findings produced several major insights:




2
People were less willing to discuss the Snowden-NSA story in social media
than they were in person. 86% of Americans were willing to have an in-person
conversation about the surveillance program, but just 42% of Facebook and Twitter users
were willing to post about it on those platforms.
Social media did not provide an alternative discussion platform for those who
were not willing to discuss the Snowden-NSA story. Of the 14% of Americans
unwilling to discuss the Snowden-NSA story in person with others, only 0.3% were willing
to post about it on social media.
In both personal settings and online settings, people were more willing to
share their views if they thought their audience agreed with them. For instance,
at work, those who felt their coworkers agreed with their opinion were about three times
more likely to say they would join a workplace conversation about the Snowden-NSA
situation.
Previous ‘spiral of silence’ findings as to people’s willingness to speak up in
various settings also apply to social media users. Those who use Facebook were
Noelle-Neumann, E. (1974). "The Spiral of Silence A Theory of Public Opinion." Journal of Communication 24(2): 43-51.
The survey was conducted between August 7-September 16, 2013 and has a margin of error of plus or minus 2.6 percentage points for the
full sample.
3
www.pewresearch.org
4
PEW RESEARCH CENTER

more willing to share their views if they thought their followers agreed with them. If a
person felt that people in their Facebook network agreed with their opinion about the
Snowden-NSA issue, they were about twice as likely to join a discussion on Facebook about
this issue.
Facebook and Twitter users were also less likely to share their opinions in
many face-to-face settings. This was especially true if they did not feel that
their Facebook friends or Twitter followers agreed with their point of view. For
instance, the average Facebook user (someone who uses the site a few times per day) was
half as likely as other people to say they would be willing to voice their opinion with friends
at a restaurant. If they felt that their online Facebook network agreed with their views on
this issue, their willingness to speak out in a face-to-face discussion with friends was
higher, although they were still only 0.74 times as likely to voice their opinion as other
people.
Overall, the findings indicate that in the Snowden case, social media did not provide new forums
for those who might otherwise remain silent to express their opinions and debate issues. Further,
if people thought their friends and followers in social media disagreed with them, they were less
likely to say they would state their views on the Snowden-NSA story online and in other contexts,
such as gatherings of friends, neighbors, or co-workers. This suggests a spiral of silence might spill
over from online contexts to in-person contexts, though our data cannot definitively demonstrate
this causation. It also might mean that the broad awareness social media users have of their
networks might make them more hesitant to speak up because they are especially tuned into the
opinions of those around them.
A rundown of the key survey findings:
People reported being less willing to discuss the Snowden-NSA story in social media than
they were in person—and social media did not provide an alternative outlet for those
reluctant to discuss the issues in person.
Fully 86% of Americans reported in the Pew Research survey they were “very” or “somewhat”
willing to have a conversation about the government’s surveillance program in at least one of the
physical settings we queried—at a public meeting, at a family dinner, at a restaurant with friends,
or at work. Yet, only 42% of those who use Facebook or Twitter were willing to discuss these same
issues through social media.
www.pewresearch.org
5
PEW RESEARCH CENTER
If the topic of the government surveillance programs came up in these settings, how
willing would you be to join in the conversation?
% of population
Very willing
Somewhat willing
At a family dinner
40
At a restaurant with friends
27
At work
27
On Twitter (based on Twitter
users)
16
14
Very unwilling
35
32
At a community meeting
On Facebook (based on FB
users)
Somewhat unwilling
12
39
41
14
16
39
16
16
27
27
14
19
23
19
13
35
40
Source, Pew Research Center Internet Project Survey August 7-September 16, 2013. N=1,801 adults.
PEW RESEARCH CENTER
Of the 14% of Americans who were not willing to discuss this issue in person, almost none (0.3%)
said they were willing to have a conversation about this issue through social media. This
challenges the notion that social media spaces might be considered useful venues for people
sharing views they would not otherwise express when they are in the physical presence of others.
Not only were social media sites not an alternative forum for discussion, social media
users were less willing to share their opinions in face-to-face settings.
We also did statistical modeling allowing us to more fully understand the findings by controlling
for such things as gender, age, education levels, race, and marital status—all of which are related to
whether people use social media and how they use it. That modeling allowed us to calculate how
likely people were to be willing to express their views in these differing settings holding other
things constant.4
We report the odds based on a logistic regression. The outcome of a logistic regression tells us the probability that a person will do
something based on the relationship to a series of predictor variables. For example, if half of the people in our sample are willing to speak out
at a public meeting, but half are not, the probability of doing something is 50%., i.e., a 50-50 percent chance, the odds are equal, 1 to 1. The
odds are a ratio of the probability that a person will do something over the probability that they will not. Let’s say, hypothetically, that 80% of
the people in our sample were willing to speak with family about an issue, this means that 20% were not. The odds that they would speak out
are .8/.2 = 4. That is to say, the odds that someone would speak with family are 4 to 1, or are 4 times higher, or are 4 times more likely to
occur. Throughout this report, we use that language.
4
www.pewresearch.org
6
PEW RESEARCH CENTER
The results of our analyses show that, even holding other factors such as age constant, social
media users are less likely than others to say they would join a discussion about the Snowden-NSA
revelations.


The typical Facebook user—someone who logs onto the site a few times per day—is half as
likely to be willing to have a discussion about the Snowden-NSA issues at a physical public
meeting as a non-Facebook user.
Similarly, the typical Twitter user—someone who uses the site a few times per day—is 0.24
times less likely to be willing to share their opinions in the workplace as an internet user
who does not use Twitter.
In both offline and online settings, people said they were more willing to share their views
on the Snowden-NSA revelations if they thought their audience agreed with them.
Previous research has shown that when people decide whether to speak out about an issue, they
rely on reference groups—friendships and community ties—to weigh their opinion relative to their
peers. In the survey, we asked respondents about their sense of whether different groups of people
in their lives agreed or disagreed with their positions on the Snowden leaks. There was some
notable variance between those who feel they know the views of their peers and those who do not
know what others think. Generally, the more socially close people were—e.g. spouses or family
members—the more likely it was that the respondents felt their views matched.
www.pewresearch.org
7
PEW RESEARCH CENTER
To what extent do you think others agree with your views about the Snowden-NSA
revelations?
% of population who say various people agree or disagree with their views
Mostly agree
Somewhat agree
Your spouse or partner
Somewhat disagree
37
Other family members
36
33
7
37
24
40
The people in your network on
Facebook (based on FB users)
24
39
20
18
10
35
Your coworkers (based on
those employed)
Your neighbors
Don’t know
53
Your close friends
The people who follow you on
Twitter (based on Twitter users)
Mostly disagree
12
11
36
29
11
11
9
7
5
12
6
6
4 4
13
19
5
21
9
25
37
Source, Pew Research Center Internet Project Survey August 7-September 16, 2013. N=1,801 adults
PEW RESEARCH CENTER
We again calculated how likely it was that someone would be willing to share their views in
different settings, depending on their sense of whether their audience agreed with them. We found
that, in the case of Snowden’s revelations about the NSA, it was clear that if people felt their
audience supported them, they were more likely to say they would join a conversation:




At work, those who felt their coworkers agreed with their opinion on the government’s
surveillance program were 2.92 times more likely to say they would join a conversation on
the topic of Snowden-NSA.
At a family dinner, those who felt that family members agreed with their opinion were 1.90
times more likely to be willing to discuss the Snowden-NSA issue.
At a restaurant with friends, if their close friends agreed with their opinion people were
1.42 times more likely to be willing to discuss the Snowden-NSA matter.
On Facebook, if a person felt that people in their Facebook network agreed with their
position on that issue, they were 1.91 times more likely to be willing to join a conversation
on the topic of Snowden-NSA.
www.pewresearch.org
8
PEW RESEARCH CENTER
Those who do not feel that their Facebook friends or Twitter followers agree with their
opinion are more likely to self-censor their views on the Snowden-NSA story in many
circumstances—in social media and in face-to-face encounters.
In this survey on the Snowden-NSA matter, we found that when social media users felt their
opinions were not supported online, they were less likely to say they would speak their minds. This
was true not only in social media spaces, but also in the physical presence of others.


The average Facebook user (someone who uses the site a few times per day) was half as
likely as other people to say they would be willing to voice their opinion with friends at a
restaurant. If they felt that their online Facebook network agreed with their views on this
issue, their willingness to speak out in a face-to-face discussion with friends was higher,
although they were still only 0.74 times as likely to voice their opinion.
The typical Twitter user (who uses the site a few times per day) is 0.24 times as likely to
share their opinions with colleagues at work as an internet user who does not use Twitter.
However, Twitter users who felt that their online Twitter followers shared their opinion
were less reserved: They were only 0.66 times less likely to speak up than other internet
users.
The survey did not directly explore why people might remain silent if they felt that their opinions
were in the minority. The traditional view of the spiral of silence is that people choose not to speak
out for fear of isolation. Other Pew Research studies have found that it is common for social media
users to be mistaken about their friends’ beliefs and to be surprised once they discover their
friends’ actual views via social media. Thus, it might be the case that people do not want to disclose
their minority views for fear of disappointing their friends, getting into fruitless arguments, or
losing them entirely. Some people may prefer not to share their views on social media because
their posts persist and can be found later—perhaps by prospective employers or others with high
status. As to why the absence of agreement on social media platforms spills over into a spiral of
silence in physical settings, we speculate that social media users may have witnessed those with
minority opinions experiencing ostracism, ridicule or bullying online, and that this might increase
the perceived risk of opinion sharing in other settings.
People also say they would speak up, or stay silent, under specific conditions.
In addition to exploring the impact of agreement/disagreement on whether people were willing to
discuss the Snowden-NSA revelations, we asked about other factors that might shape whether
people would speak out, even if they suspected they held minority views. This survey shows how
the social and political climate in which people share opinions depends on several other things:

Their confidence in how much they know. Those who felt they knew a lot about the
issues were more likely than others to say they would join conversations.

The intensity of their opinions. Those who said they had strong feelings about the
Snowden-NSA matter were more willing than those with less intense feelings to talk about
the subject.
www.pewresearch.org
9
PEW RESEARCH CENTER

Their level of interest. Those who said they were very interested in the Snowden-NSA
story were more likely than those who were not as interested to express their opinions.
People’s use of social media did little to increase their access to information about the
Snowden-NSA revelations.
We asked respondents where they were getting information about the debates swirling around the
Snowden revelations, and found that social media was not a common source of news for most
Americans. Traditional broadcast news sources were by far the most common sources. In contrast,
social media sources like Facebook and Twitter were the least commonly identified sources for
news on this issue.






58% of all adults got at least some information on the topic of Snowden-NSA from TV or
radio.
34% got at least some information from online sources other than social media.5
31% got at least some information from friends and family.
19% got at least some information from a print newspaper.
15% got at least some information while on Facebook.
3% got at least some information from Twitter.
There are limits to what this snapshot can tell us about how social media use is related to the ways
Americans discuss important political issues. This study focuses on one specific public affairs issue
that was of interest to most Americans: the Snowden-NSA revelations. It is not an exhaustive
review of all public policy issues and the way they are discussed in social media.
The context of the Snowden-NSA story may also have made it somewhat different from other
kinds of public debates. At the time of this study, the material leaked by Edward Snowden related
to NSA monitoring of communications dealt specifically with “meta-data” collected on people’s
phone and internet communications. For a phone call, the meta-data collected by the NSA was
described as including the duration of the call, when it happened, the numbers the call was
between, but not a recording of the call. For email, meta-data would have included the sender and
recipient’s email addresses and when it was sent, but not the subject or text of the email.
Additional information leaked by Snowden after our study was completed suggests that Western
intelligence agencies monitored and manipulated the content of online discussions and the NSA
recorded the content of foreign phone calls. In reaction to these additional revelations, people may
have adjusted their use of social media and their willingness to discuss a range of topics, including
public issues such as government surveillance. However, given the limited extent of the
information leaked by Snowden at the time the survey was fielded, it seems unlikely that the
average American had extensively altered their willingness to discuss political issues. Future
research may provide insight into whether Americans have become more or less willing to discuss
specific issues on-and offline as a result of government surveillance programs. While this study
In this survey, 80% of adults said they were internet users, 71% of the internet users are Facebook users, and 18% of internet users are
Twitter users.
5
www.pewresearch.org
10
PEW RESEARCH CENTER
focused on the Snowden-NSA revelations, we suspect that Americans use social media in similar
ways to discuss and get news about other political issues.
About this survey
This report contains findings from a nationally representative survey of 1,801 American adults
(ages 18+) conducted by the Pew Research Center and fielded August 7-September 16, 2013 by
Princeton Research Associates International. It was conducted in English and Spanish on landline
(N=901) and cell phones (N=900). The margin of error for the full sample is plus or minus 2.6
percentage points. Some 1,076 respondents are users of social networking sites and the margin of
error for that subgroup is plus or minus 3.3 percentage points.
www.pewresearch.org
11
PEW RESEARCH CENTER
Table of Contents
About this Report
1
About the Pew Research Center
2
Summary of Findings
3
Main Analysis: Political Issues and the Spiral of Silence
12
Where people got news about the Snowden-NSA surveillance story
12
People’s overall willingness to share their views
14
Appendix
27
Survey Questions
31
Methods
38
www.pewresearch.org
12
PEW RESEARCH CENTER
Main Analysis: Political Issues and the Spiral of Silence
It has long been clear in the research community that people’s willingness to discuss political
issues depends on their access to news and on the social climate for discussion. This study explores
people’s willingness to share their opinions on and offline about an important political issue. The
report is built on Pew Research Center survey findings related to how people use social media, as
well as traditional media, to get information on one political issue that dominated the news in the
summer of 2013: the revelations by defense contractor Edward Snowden. In June 2013, Snowden
leaked classified documents to The Washington Post and Britain’s Guardian newspaper about
surveillance by the U.S. National Security Agency and some allied governments into the phone
calling records and email exchanges of untold numbers of persons.6
We asked people where they were getting information about the debates swirling around the
Snowden revelations, and found that social media was not a common source of news for most
Americans. Traditional broadcast news sources were by far the most common sources. In contrast,
social media sources (Facebook and Twitter) were the least commonly identified sources for news
on this issue.






58% of all adults got at least some information about this topic from TV or radio.
34% got at least some information from online sources other than social media.7
31% got at least some information from friends and family.
19% got at least some information from a print newspaper.
15% got at least some information while on Facebook.
3% got at least some information from Twitter.
Looking only at those Americans who use either Facebook or Twitter, 26% of Facebook users and
22% of Twitter users reported being exposed to at least some information about the government’s
surveillance program on these platforms.
A relatively small number of Americans—12%—reported receiving no information about the
debates over the government’s telephone and digital surveillance program. Some 15% of
Americans said they relied on a single source of information about this issue. The majority relied
on at least three information sources.
This reported use of Facebook and Twitter for news about the Snowden revelations is
substantively lower than what has been reported previously for use of these platforms to access
news more broadly. Data from the Pew Research Center’s (2013) report on “News Use across
Social Media Platforms,” conducted over the same time period as our survey, found that 47% of
We also asked about people’s use of mobile devices, Facebook, Twitter, Instagram, Pinterest, and LinkedIn and that material is a core part
of the analysis. In this survey, 80% of adults say they are internet users and 89% said they have cell phones. A detailed demographic
breakdown of the demographics of users of various social media platforms in this survey can be found here.
7 In this survey, 80% of adults said they were internet users, 71% of the internet users are Facebook users, and 18% of internet users are
Twitter users.
6
www.pewresearch.org
13
PEW RESEARCH CENTER
Facebook users and 52% of Twitter users use these platforms to consume news. One explanation
for the difference in our findings likely relates to the fact that in this survey we were asking about a
single public issue, while the other Pew Research survey included broader types of news, including
entertainment, sports, and politics.8



Some might expect that internet users in general and social media users in particular are
less likely to rely on traditional media sources for news on political issues because they
have alternative sources. But, for internet users in general, and for most social media users,
we find the opposite to be true. Using regression analysis to control for demographic
characteristics, we find: Internet users are more likely than non-users to get news
on the surveillance story from TV and radio. An internet user is 1.63 times more
likely to have obtained even a little news on the Snowden-NSA revelations from radio and
television than a non-internet user.
Twitter users are more likely than non-Twitter users to get news on the
surveillance story from TV and radio. A typical Twitter user (someone who uses the
site a few times per day) is 2.25 times more likely to have obtained news on this issue
through TV and radio than an internet user who does not use this platform, and 3.67 times
more likely than a non-internet user.
Instagram users were also more likely to get news on the surveillance story
from traditional broadcast sources. A typical user of Instagram (someone who uses
the site a few times per day) was 2.46 times more likely to have received television and
radio news on this topic in comparison with an internet users who does not use Instagram,
and 4.02 times more likely than a non-internet user.
This contrasts with the situation that applies to users of some other social media platforms:


The typical Pinterest user (who uses the site a couple of times per week) is 0.92 times less
likely to get news about the government’s surveillance program from TV and radio in
comparison with an internet user who does not use this platform, but he or she is still 1.51
times more likely to get news from TV and radio than a non-internet user.
Similarly, someone who uses LinkedIn a couple of times per week is 0.87 times less likely
to get news on this issue from television and radio compared to an internet user who does
not use LinkedIn, but still 1.41 times higher than for a non-internet user.
Facebook users are no more or less likely to obtain news through TV and radio than other internet
users.
While some social media do seem to distract from traditional media sources, on the whole, these
effects are relatively small. Someone who uses multiple social media sites at a typical level of use—
Facebook, Twitter, Instagram, Pinterest, and LinkedIn—is about 8 times more likely than nonSee for instance: “Low marks for the 2012 election” available at: http://www.people-press.org/2012/11/15/section-4-news-sourceselection-night-and-views-of-press-coverage/
Also: “Internet Gains Most as Campaign News Source but Cable TV Still Leads” available at: http://www.journalism.org/2012/10/25/socialmedia-doubles-remains-limited/
Also: “Internet Gains on Television as Public’s Main News Source” available at: http://www.people-press.org/2011/01/04/internet-gains-ontelevision-as-publics-main-news-source/.
8
www.pewresearch.org
14
PEW RESEARCH CENTER
internet users and 5 times more likely than internet users who do not use social media to get
information about the government’s surveillance program through TV and radio (See Appendix,
Table A).9 For the most part, social media users did not get their news through social media, they
got it through television and radio.
Controlling for other factors, internet and social media use do not account for any of the difference
in use of print newspapers to find information on the topic of the government’s surveillance
program. Internet users, including those who use Facebook, Twitter, Instagram, Pinterest, and
LinkedIn, are as likely as anyone else to use newspapers for news about the government’s
surveillance program.
Social media did not provide an alternative outlet for the 14% of Americans who were not
willing to discuss the Snowden-NSA issue in person
While it has been suggested that social media might provide new channels for communication
about important political issues, our survey suggests that few people are willing to deliberate
online who would not also do so in person. Almost everyone in our sample who reported that they
would be willing to discuss something on Twitter or Facebook also indicated that they would be
willing to have a conversation on this topic in an offline setting. Only 0.3% of Americans reported
that they were not willing to have a conversation about the government surveillance program
when people were physically present, but were willing to have such a conversation through social
media.
There are many social situations where people might have the opportunity to discuss political
issues. We asked respondents to tell us how willing they would be to join a conversation “if the
topic of the government’s surveillance programs came up” in a variety of settings, online and
offline. We asked them how willing they would be to join in the conversation at a community
meeting, at work, at a restaurant with friends, at a family dinner, on Facebook, and on Twitter.
In most social settings, the majority of Americans reported that they would be willing to join a
conversation about the Snowden-NSA revelations. The only settings where most people were not
willing to discuss their opinion was on Facebook and Twitter.

74% of all adults said they would be “very” or “somewhat” willing to join the conversation if
the Snowden-NSA story came up at a family dinner.

74% of all adults said they would be “very” or “somewhat” willing to join the conversation if
the Snowden-NSA story came up at a restaurant with friends.

66% of all adults said they would be “very” or “somewhat” willing to join the conversation if
the Snowden-NSA story came up at a community meeting.
Typical use of the social media in this example includes logging into the Twitter and Instagram platforms a few times a day, using Pinterest a
half dozen times per month, and using LinkedIn a couple of times per month (frequency of Facebook use was not statistically significant).
9
www.pewresearch.org
15
PEW RESEARCH CENTER

65% of employed adults said they would be “very” or “somewhat” willing to join the
conversation if the Snowden-NSA story came up at work.

42% of Facebook users said they would be “very” or “somewhat” willing to join the
conversation on Facebook. 10

41% of Twitter users said they would be “very” or “somewhat” willing to join the
conversation if the Snowden-NSA story came up on Twitter.
If the topic of the government surveillance programs came up in these
settings, how willing would you be to join in the conversation?
% of population
Very willing
Somewhat willing
At a family dinner
40
At a restaurant with friends
27
At work
27
On Twitter (based on Twitter
users)
16
14
Very unwilling
35
32
At a community meeting
On Facebook (based on FB
users)
Somewhat unwilling
12
39
41
14
16
39
16
16
27
27
14
19
23
19
13
35
40
Source, Pew Research Center Internet Project Survey August 7-September 16, 2013. N=1,801 adults.
PEW RESEARCH CENTER
In all, 86% of Americans were willing to have a conversation in the physical presence of others—
that is, at a public meeting, at a family dinner, at a restaurant with friends, or at work on the topic
of the government’s surveillance program. Only 42% of those who use Facebook and 41% of
Twitter users felt comfortable discussing this same issue through social media.
10
This figure does not match the figure in the chart below because of rounding.
www.pewresearch.org
16
PEW RESEARCH CENTER
Exploring the conditions under which people are willing to speak
Previous research showed that when people decide whether to speak out about an issue, they rely
on reference groups—friendships and community ties—and weigh their opinion relative to these
groups before speaking out in a setting. Other factors also play a role in people’s willingness to
discuss issues. Our survey found that if people had a strong interest in the topic of the SnowdenNSA revelations, held a strong opinion, and felt knowledgeable about it, they were generally more
willing to join a conversation about this issue.
What follows is our detailed exploration of the various circumstances that might affect someone’s
willingness to speak about issues—in this case, the Snowden-NSA revelations.
In most settings, people’s level of interest in the Snowden-NSA revelations was related to
their willingness to discuss this topic
In the summer of 2013, interest in information leaked by Edward Snowden about the U.S.
government’s telephone and digital surveillance
programs was high.
In our survey, respondents were asked how
interested they were in debates about “a
government program with the aim of collecting
information about people’s telephone calls,
emails, and other online communications.” Some
60% of American adults reported they were very
or somewhat interested in this topic. Only 20%
of Americans reported that they were not
interested at all.
Using regression analysis to control for
demographic differences, we found that someone
who was “very interested” in the government
surveillance program was 1.78 times more likely
to be willing to join a conversation at a
community meeting than someone who has no
interest at all (See Appendix, Table B).
Similarly, compared with someone who was
uninterested in this topic, someone who was very
interested was 2.64 times more likely to speak up
during a conversation with friends at a
restaurant, and 2.88 times more likely to speak
up when talking with family at dinner.
The level of interest in the NSA
surveillance story was relatively high
% of adults who say they were interested in the
revelations about government surveillance programs
40
34
30
26
20
19
20
Not too
interested
Not interested
at all
10
0
Very interested
Somewhat
interested
Source, Pew Research Center Internet Project Survey August 7September 16, 2013. N=1,801 adults
PEW RESEARCH CENTER
People’s level of interest in the Snowden-NSA story was not related to willingness to speak up in
the workplace or on Facebook. The regression analyses showed that the interested and the
www.pewresearch.org
17
PEW RESEARCH CENTER
uninterested were equally as likely to say they would voice their opinions on Facebook and at
work. 11
Those with more fervent opinions about the
Snowden-NSA story were more likely to say
they would speak out
In addition to asking about their interest in the
Snowden-NSA story, we asked whether
respondents favored or opposed “a government
program to collect nearly all communications in
the U.S. as part of anti-terrorism efforts?” Some
37% of Americans strongly or somewhat favored
the surveillance programs and 52% strongly or
somewhat opposed them. Another 10% said they
didn’t know or refused to answer the question.12
Americans are split on whether they
support or oppose government
surveillance programs
% of adult responses to the question: Do you favor or
oppose a government program to collect nearly all
communications in the U.S. as part of anti-terrorism
efforts?
40
30
30
24
22
20
We found that those who had stronger opinions
13
on the topic of the Snowden-NSA revelations
10
were more willing to speak out on this issue at
public meetings, with family over dinner, and on
Facebook (See Appendix, Table B). In
0
comparison with those with less intense
Strongly
Somewhat Somewhat
Strongly
favor
favor
oppose
oppose
opinions, someone who either “strongly” favors
or opposes the collection of domestic
Source, Pew Research Center Internet Project Survey August 7communications as part of government
September 16, 2013. N=1,801 adults
surveillance program was 1.56 times more likely PEW RESEARCH CENTER
to be willing to speak out at a public meeting,
1.35 times more likely to be would willingly discuss the issue with family over dinner, and 2.40
times more likely to have said they would join a conversation on Facebook.
We were unable to perform a regression analysis on willingness to discuss the government’s surveillance program on Twitter because of the
small number of Twitter users in our sample, and the even smaller number who expressed a willingness to discuss public affairs on this
platform.
12 When designing a question about the government’s surveillance program, we based the question on the findings of the Pew Research
report on “Government Surveillance: A Question Wording Experiment” that was conducted in the weeks prior to our survey. We designed our
question with the intent that roughly equal numbers of Americans would agree and disagree. It was important to pick a topic about which
Americans were likely to be split in their opinions, so that we could observe variation in people’s behavior as they considered sharing their
views with those around them.
11
www.pewresearch.org
18
PEW RESEARCH CENTER
Those who felt more knowledgeable were more willing to discuss the Snowden-NSA story
When a new, potentially important issue appears in the news, those who feel knowledgeable tend
to show greater willingness to have a conversation with others. Indeed, feeling knowledgeable
about this issue increased the likelihood
The level of knowledge Americans felt they
that someone would be willing to join a
had about the debates surrounding
conversation about the government’s
government surveillance programs
surveillance program in all of the settings
we explored.
% of internet users who give these answers
In this survey, participants were asked to
report on how knowledgeable they felt
about the debate surrounding “government
programs aimed at collecting information
about people’s calls, emails and other
online communication.”
Some 54% of adults reported that they felt
very or somewhat knowledgeable about the
government surveillance programs and
45% said they felt they had little or no
knowledge of this topic.
50
42
40
28
30
17
20
12
10
0
Very
Somewhat
Not too
Not
knowledgeable knowledgeable knowledgeable knowledgeable
at all
Compared with someone who did not feel
that they had any knowledge about the
Source, Pew Research Center Internet Project Survey August 7-September
topic, those who described themselves as
16, 2013. N=1,801 adults
“very knowledgeable” were 2.68 times
PEW RESEARCH CENTER
more likely to join a conversation at a
public meeting, 3.19 times more likely in the workplace, 2.01 times more likely with friends at a
restaurant, 1.79 times more likely over dinner with family, and 2.36 times more likely on Facebook
(See Appendix, Table B).
People’s awareness of the opinions of those around them: Those who use social media
tend to be more aware of others’ views
The level of awareness that people have of other people’s opinions plays a significant role in how
willing they are to share their opinions. It has long been established that when people are
surrounded by those who are likely to disagree with their opinion, they are more likely to selfcensor.
We examined the awareness that people felt they had about the opinions of family, friends,
coworkers, and others about the Snowden-NSA story—and the degree to which people think these
other connections agree or disagree with them. We find that people were most likely to say they
were aware of others’ views when it involved a very close relationship, such as a spouse/partner or
close friends. Fully 96% of those who are married or living with a partner believe they know their
spouse’s/partner’s opinion on the topic of the government’s surveillance program.
www.pewresearch.org
19
PEW RESEARCH CENTER
For other kinds of relationships, though, there was more variance in respondents’ answers.





96% of people who are married or living with a partner report that they know their
partner’s opinion.
88% of people reported knowing the opinions of their close friends.
87% of people feel they know the opinions of their family members.
80% of people who are employed reported knowing the opinions of their coworkers.
62% of people feel they know their neighbors’ opinions on this issue.
The awareness that people have of the opinions of their followers on social media tends to be lower
than for most other types of relationships.


Of Facebook users, 76% felt they knew the opinions of people in their network.
Of Twitter users, 68% felt they knew the opinions of those who followed them.
Interestingly enough, social media users are more likely than others to report they are aware of the
opinions of different people in their lives.



93% of Twitter users and 90% of Facebook users say they know the opinions of family
members on the Snowden-NSA issue. This compares with 82% of non-internet users and
84% of internet users who do not use social media.
94% of Twitter users and 91% of Facebook users say they are aware of their close friends’
opinions on the Snowden-NSA topic. This compares with 82% of non-internet users and
85% of internet users who do not use these social media sites.
66% of Facebook users, and 71% of Twitter users say they know their neighbors’ opinions
about the government’s surveillance programs. This compares with 60% of internet users
who are not social media users.
www.pewresearch.org
20
PEW RESEARCH CENTER
Social media users were more likely than other internet users to say they knew the
views of those around them about the Snowden-NSA story
% who say they know the opinions of others
Twitter users
100
94
91
Facebook users
85
93
Internet users who don't use social media
90
84
80
71
60
66
60
40
20
0
Know views of close friends
Know views of family members
(not spouse/partner)
Know views of neighbors
Source, Pew Research Center Internet Project Survey August 7-September 16, 2013. N=1,801 adults.
PEW RESEARCH CENTER
The more social media platforms people use, the greater their awareness of opinions in their
extended network. When asked to report on the opinions of the people in their Facebook network,
79% of Facebook users say they know the opinions of their Facebook friends. Of those who use
Twitter and Facebook, 86% say they know the opinions of their Facebook friends.
One exception to the trend of internet users knowing more about those in their social networks is
coworkers. Employed non-internet users tend to be a bit more aware of colleagues’ opinions than
internet users. Some 85% of employed non-internet users say they are aware of their coworkers’
opinions, compared with 78% of internet users who do not use social media, 82% of Facebook
users, and 84% of Twitter users who say they know the opinions of coworkers.
How much people think they agree with the views of family members, friends, and
colleagues
A crucial issue affecting whether someone will be willing to discuss a controversial subject is the
degree to which a speaker feels his or her views line up with their audience. Some research has
found that people have a tendency to associate with those who share their opinions. That is, even
though broad public opinion may be divided on an issue, people are more likely to believe that
their acquaintances support their position on that issue. Some of this similarity is a result of
homophily, the penchant for people to associate with people like themselves; some is a result of
www.pewresearch.org
21
PEW RESEARCH CENTER
the influence of opinion leaders; and some is a result of the tendency for people to assume more
agreement than there is in reality.13
This survey shows that people have different notions about how much agreement they have with
close family and friends, compared with associates that are less close, including those in their
Facebook and Twitter networks. In addition, the more socially distant an audience is, the more
likely it is that respondents did not know the views of their potential audience.

86% of those who are married or living with a partner believe their spouse’s/partner’s
views “mostly” or “somewhat” agree with theirs about the Snowden-NSA revelations.

74% of all adults believe their close friends “mostly” or “somewhat” agree with their
views about the Snowden-NSA revelations.

70% of all adults believe their family members “mostly” or “somewhat” agree with their
views about the Snowden-NSA revelations. (This includes family members who are not a
spouse or partner.)

64% of those who are employed think that their coworkers agree with their position on the
government’s surveillance program.

63% of Facebook users believe that the people in their Facebook network “mostly” or
“somewhat” agree with their views about the Snowden-NSA revelations.

56% of Twitter users believe that the people who follow them on Twitter “mostly” or
“somewhat” agree with their views.

47% of all adults believe their neighbors “mostly or “somewhat” agree with their views.
Goel, S., W. Mason, et al. (2010). "Real and Perceived Attitude Agreement in Social Networks." Journal of Personality and Social Psychology
99(4): 611-621.
13
www.pewresearch.org
22
PEW RESEARCH CENTER
To what extent do you think others agree with your views about the Snowden-NSA
revelations?
% of population who say various people agree or disagree with their views
Mostly agree
Somewhat agree
Your spouse or partner
Somewhat disagree
37
Other family members
36
33
7
37
24
40
The people in your network on
Facebook (based on FB users)
24
39
20
18
10
35
Your coworkers (based on
those employed)
Your neighbors
Don’t know
53
Your close friends
The people who follow you on
Twitter (based on Twitter users)
Mostly disagree
12
11
36
29
11
11
9
7
5
12
6
6
4 4
13
19
5
21
9
25
37
Source, Pew Research Center Internet Project Survey August 7-September 16, 2013. N=1,801 adults
PEW RESEARCH CENTER
Facebook users were more likely to feel that friends, family and acquaintances share their opinion.
Regression analysis was used to control for demographic characteristics, interest in the topic,
knowledge of the topic, strength of opinion on this issue, and social media use when predicting
agreement with different types of acquaintances. We find that Facebook use is related to perceived
agreement with the opinions of friends, family, and other acquaintances (See Appendix, Table C).
Users who contribute content and read other people’s content on Facebook are more likely to
believe that other people agree with their opinions.


Someone who frequently uses the “like button” on Facebook content contributed by other
Facebook users (they use it a few times per day) is 1.88 times more likely to feel that their
family members share their views, and they are 1.72 times more likely to feel they share the
opinions of people in their Facebook network, when compared to those who do not use the
like button.
Someone who updates their status on Facebook a half dozen times per month, compared to
someone who does not update at all, is 1.10 times more likely to feel they share the
opinions of family members, and 1.13 times more likely to share the opinions of their close
friends.
www.pewresearch.org
23
PEW RESEARCH CENTER
It is not immediately clear from our study why Facebook activities are related to perceptions of
higher levels of agreement with Facebook friends. Two possible explanations are related to
“cyberbalkanization.”14 Facebook friendship networks may be more likely to consist of similar
people, or their opinions may become more similar over time. However, we expect that a third
option is most likely. Reading content contributed by other users, actively clicking the like button,
as well as receiving feedback in response to status updates, provides for enhanced observation of
others and confirmatory feedback from friends and family. In addition to people choosing to
associate with people on Facebook who are similar to them, Facebook makes people more aware of
existing opinion similarity.
The spiral of silence persists online and offline: People are less likely to speak when they
think their audiences disagree with them
In many settings, it is not well understood how much people self-censor in response to such social
pressures. Some early research has shown that the rate of self-censorship on Facebook is very
high. One study found that people on Facebook start to write, but ultimately fail to share, 33% of
posts and 13% of comments.15 This self-censorship has been described as a response to “context
collapse”16—that is, people deciding not to share content that is of personal interest, but is unlikely
to appeal to a social media audience that focuses on narrow topics.
However, there is another possibility. Some self-censorship might be the result of feeling that
social media followers are likely to object or disagree with their opinion. In other words, a user
might know the content is relevant to some followers, but decide not to share it on social media for
fear of inviting disagreement among their followers.




At work, those who felt their coworkers agreed with their opinion were 2.92 times more
likely to say they would join a conversation on the Snowden-NSA topic than for those who
did not feel they would agree with their coworkers’ opinion on the government’s
surveillance program.
At a family dinner, those who felt that family members agreed with their opinion were 1.90
times more likely to speak out about Snowden-NSA issue.
At a restaurant with friends, those who felt that their close friends agreed with their
opinion were 1.42 times more likely to share their opinions.
On Facebook, if the person felt that people in their Facebook network agreed with their
position on this issue, they were 1.91 times more likely to join a conversation about
Snowden-NSA.
However, the social pressure from some types of relationships carried across multiple settings. For
example, when at a restaurant with friends, people’s willingness to speak out was tied to the
14
Sunstein, C. R. (2001). Republic.com. Princeton, N.J., Princeton University Press.
15
Das, S. and A. Kramer (2013). "Self-censorship on Facebook." Proc. of ICWSM 2013: 120-127.
Marwick, A. E. and d. boyd (2010). "I Tweet Honestly, I Tweet Passionately: Twitter Users, Context Collapse, and the Imagined Audience."
New Media & Society 13(1): 114-133.
16
www.pewresearch.org
24
PEW RESEARCH CENTER
opinions of their family members. That might possibly be the case because close friends and family
tend to have similar opinions. Or it might be the case because a meal with friends at a restaurant
may include family. Additionally, it might be the case because people felt they knew they had
supportive family members kind of “standing by” them. Whatever the reason, those who had
family that shared their opinions were 1.42 times more likely to join a conversation about this
issue at a restaurant with friends, even when friends did not agree.
When social media followers disagree, people are more likely to self-censor offline
In some offline settings, we found that when compared to non-internet users, online Americans in
general were more willing to join a conversation about the Snowden-NSA story. An internet user
was 2.41 times more likely to be willing to have a conversation at work, and 1.49 times more likely
to have a conversation with family about the government’s surveillance program. A typical
LinkedIn user, who accesses the site a half dozen times per month, was 1.20 times more likely to
discuss this political issue in a restaurant with friends than other internet users or non-internet
users.
However, we found many more examples to suggest that social media use is associated with a
lower likelihood that people would have a conversation on a political issue in physical settings.
When controlling for demographic traits such as gender, age, race, educational attainment, and
marital status, as well as variation in interest, opinion strength, knowledge, and other sources of
information exposure we found:
Facebook users were less willing to discuss the government’s surveillance program
at a public meeting. Someone who uses Facebook several times per day is 0.53 times less likely
to be willing to discuss the Snowden-NSA topic at a public meeting than someone who does not
use the Facebook platform at all.
Instagam users were less likely to say they would discuss the government’s
surveillance program at a family dinner or at a restaurant with friends. A typical
Instagram user (who uses the platform several times per day) is 0.49 times less likely to be willing
to join a conversation about the government’s surveillance program with family at dinner, and
0.44 times less likely with friends at a restaurant, than for people who do not use Instagram.
It is not completely clear why some users of social media would be less willing to share an opinion
in physical settings. However, since we have controlled for demographic differences, and variation
in interest, opinion strength, knowledge, and other sources of information exposure, it is possible
that this heightened self-censorship might be tied to social media users’ greater awareness of the
opinions of others in their network (on this and other topics). Thus, they could be more aware of
views that oppose their own.
If their use of social media gives them broader exposure to the views of friends, family, and
workmates, this might increase the likelihood that people will choose to withhold their opinion
because they know more about the people who will object to it.
www.pewresearch.org
25
PEW RESEARCH CENTER
There are two additional examples from our data that most clearly demonstrate this relationship.
Twitter users were less willing to engage in a conversation in the workplace,
especially if they felt those following them on Twitter did not agree with their
opinion on the government’s surveillance program. A typical Twitter user, who uses the
platform several times per day, was0.24 times less likely to be willing to join a conversation on the
Snowden-NSA story at work than other internet users. However, if they felt their Twitter followers
agreed with their opinion, then they were only 0.69 times less likely to be willing to engage in a
discussion at work. This relationship was in addition to the lower likelihood that someone would
speak out at work if they felt their coworkers did not share their opinions.
Facebook users and those who do not feel their Facebook friends agree with their
opinion were less willing to engage in an in-person discussion with friends on this
issue. A typical Facebook user, someone who accesses the platform several times per day, is 0.53
times less likely to be willing to discuss the government’s surveillance program with friends at a
restaurant than those who do not use Facebook. If they feel that people in their Facebook network
agree with their opinion, they are only 0.74 times less likely to discuss this topic in-person with
friends when compared with those who do not use Facebook at all. This relationship is in addition
to the lower likelihood that people have of speaking out when at a restaurant if they do not believe
their close friends agree with their opinion. Facebook likely increases awareness of the diversity of
opinions in people’s friendship network beyond their closest friends. This awareness reduces
certainty in the similarity of opinions between friends and increases the fear of isolation or
ostracism that might result from sharing a divergent point of view.
Social media use does encourage more discussion among some groups
While social media use may be linked to a muting effect on discussions of political issues in some
physical settings, for some it is associated with new opportunities for discussion.
Unsurprisingly, the heaviest users of Facebook, in terms of frequency of commenting and private
messaging, were also those who were most likely to be willing to discuss the government’s
surveillance program on the Facebook platform. However, for all but the most intensive users, the
relationship to discussing political issues is relatively small. Someone who comments on other
people’s Facebook statuses, photos, links, and other content about twice per week was only 1.04
times more likely to be willing to discuss the Snowden-NSA story on Facebook in comparison with
someone who does none of these things.
One type of social media use was associated with a lower level of willingness to join a conversation
about public affairs on Facebook. Possibly as a result of the diversity they observed through images
contributed to Instagram, Instagram users were less willing than other Facebook users to use the
Facebook platform to discuss the government’s surveillance program. A typical Instagram user,
someone who uses the platform several times per day, was 0.49 times less likely to be willing to
discuss the government’s surveillance program on Facebook.
There are some indications that Facebook may democratize discussion of political issues in at least
some respects. Unlike many physical settings, on Facebook, those with fewer years of formal
www.pewresearch.org
26
PEW RESEARCH CENTER
education were the most likely to speak up about an important political issue. When discussing
political issues with friends at a restaurant, and family over dinner, it is those with the most
education who are most willing to join in on a conversation. The opposite is true on Facebook.
Those with the most years of formal education are more likely to fall silent when discussing the
Snowden-NSA issue. Someone with only a high school diploma was 1.34 times less likely to be
willing to join a conversation on Facebook about the government’s surveillance program when
compared to someone with an undergraduate university degree. Similarly, on Facebook, women
are as likely as men to feel comfortable discussing an important political issue. This contrasts with
discussions at community meetings and at work where women tend to feel less comfortable
discussing a political issue such as the government’s surveillance program.
www.pewresearch.org
27
PEW RESEARCH CENTER
Appendix: Regressions
Table A: Likelihood of getting information from
news sources—logistic regression
Independent Variables
Constant
Newspaper
(N=1763)
Radio & TV
(N=1763)
Odds Ratio
Odds Ratio
0.073 ***
0.092***
Female
0.839
0.780
Age
1.024 ***
1.033***
Education
1.064 **
1.053
Married or living with a partner
1.324 **
1.614***
Black/African-American
1.219
0.803
Internet user
0.958
1.634**
Cell phone user
0.722
1.400
Facebook visits per month (0-90)
0.996
1.002
LinkedIn visits per month (0-90)
0.996
0.982**
Twitter visits per month (0-90)
1.003
1.009*
Instagram visits per month (0-90)
1.000
1.010**
Pinterest visits per month (0-90)
1.001
0.990*
Number of Facebook friends
1.000
1.000
Status update per month (0-90)
0.998
0.995
‘Like’ per month (0-90)
1.003
0.998
Comment per month (0-90)
1.002
0.997
Sending message per month (0-90)
1.002
1.002
Interest in this topic (0-3)
1.287 ***
1.643***
R-squared (Nagelkerke)
0.099 ***
0.178 ***
Demographics
Media Use
Internet Activities
Facebook Activities
Other variables
Notes: N is smaller than 1801 (total sample size) because some
respondents did not answer questions about their demographics or
media use.
*p<.05 **p<.01 ***p<.001
www.pewresearch.org
28
PEW RESEARCH CENTER
Table B: Likelihood of being willing to join a conversation about the government
surveillance program in various contexts—logistic regression
Community
Meeting
(N=1763)
At Work
With Friends
(N=982)
(N=1763)
At Family
Dinner
(N=1763)
Odds Ratio
Odds Ratio
Odds Ratio
Odds Ratio
0.457
0.168**
0.333 *
0.151 ***
0.459
0.664**
0.806
1.079
0.801
Age
0.695**
1.002
0.999
0.986 ***
0.996
0.992
Education
0.976
0.992
1.102 ***
1.081 **
0.929 *
Married or living with a partner
1.257*
1.455*
1.070
1.097
1.158
Black/African-American
0.909
0.720
0.873
0.769
1.185
Interest in this topic (0-3)
1.211***
1.099
1.382 ***
1.422 ***
1.052
Knowledge of this topic (0-3)
1.390 ***
1.472 ***
1.262 **
1.215 *
1.331 **
Strong opinion on this topic
1.560 ***
1.112
1.227
1.353 *
2.397 ***
Newspaper (0-3)
1.114
1.137
1.002
0.971
1.148
Radio & TV (0-3)
0.999
1.067
1.025
1.166 *
0.858 *
Friends & family (0-3)
1.101
1.135
1.151 *
1.167 *
0.992
Facebook (0-3)
1.106
0.761**
0.984
0.873
1.272 **
Twitter (0-3)
0.692*
0.948
0.714
0.956
1.192
Other online source (0-3)
1.123
0.956
1.000
1.111
1.059
Internet user
1.290
2.414 **
1.269
1.485 *
Cell phone user
0.844
1.278
0.717
0.840
1.502
0.998
0.993 **
1.000
1.001
LinkedIn visits per month (0-90)
0.993**
1.010
1.013
1.030 *
1.001
0.990
Twitter visits per month (0-90)
0.995
0.985**
1.005
1.000
0.991
Instagram visits per month (0-90)
0.996
1.005
0.991 **
0.992 *
0.992 *
Pinterest visits per month (0-90)
1.005
1.011
1.007
1.011
1.004
Independent Variables
Constant
On Facebook
(N=948)
Odds Ratio
Demographics
Female
Government Surveillance Topic
Source of Information
Media Use
-
Internet Activities
Facebook visits per month (0-90)
www.pewresearch.org
29
PEW RESEARCH CENTER
Table B. (Cont.)
Independent Variables
Community
Meeting
(N=1763)
At Work
With Friends
(N=982)
(N=1763)
At Family
Dinner
(N=1763)
Odds Ratio
Odds Ratio
Odds Ratio
Odds Ratio
On Facebook
(N=948)
Odds Ratio
Facebook Activities
Number of Facebook friends
1.000
1.000
1.000
1.000
1.000
Status update per month (0-90)
1.007
1.002
1.004
0.993
1.009
‘Like’ per month (0-90)
1.003
1.000
1.001
1.005
0.999
Comment per month (0-90)
1.001
1.003
1.003
0.997
1.007 *
Sending messages per month (0-90)
1.001
0.996
0.999
1.003
1.007 *
Perceived Opinion Congruence (Agree = 1)
Family members
1.291
1.283
1.418 *
1.898 ***
0.880
Friends
1.307
0.854
1.417 *
1.182
0.662
Coworkers
1.074
2.916***
0.990
1.013
1.188
Neighbors
1.220
0.781
1.063
0.909
1.184
Facebook friends
1.167
1.095
1.390 *
0.876
1.911 ***
Twitter followers
1.480
2.805**
0.986
1.518
1.686
R-squared (Nagelkerke)
0.195 ***
0.210 ***
0.209 ***
0.230 ***
0.232 ***
Notes: N is smaller than 1801 (total sample size) because some respondents did not answer questions
about their demographics or media use; the analysis for at work is limited to participants who reported
having a full or part-time job; the analysis of Facebook is limited to participants who use Facebook.
*p<.05 **p<.01 ***p<.001
www.pewresearch.org
30
PEW RESEARCH CENTER
Table C: Likelihood of perceiving opinion congruence in various contexts—
logistic regression
Spouse
(N=999)
Independent Variables
Constant
Family
Members
(N=1763)
Friends
(N=1763)
Co-Workers Neighbors
(N=982)
(N=1417)
Facebook
(N=948)
Odds Ratio Odds Ratio Odds Ratio Odds Ratio Odds Ratio Odds Ratio
2.175
1.308
1.022
2.095
1.348
1.460
Female
0.705
0.916
0.741 **
0.601 ***
0.835
0.992
Age
0.999
0.990**
0.990 **
0.990
0.983 ***
0.975 ***
Education
1.006
1.023
1.041
0.977
0.997
0.984
-
1.110
1.115
1.092
1.368 **
1.092
0.297***
0.675*
0.788
0.974
1.436 *
0.956
Interest in this topic (0-3)
1.002
1.097
1.213 ***
1.154 *
1.096
1.119
Knowledge of this topic (0-3)
1.275 *
1.221 **
1.153 *
1.031
1.019
1.043
Strong opinion on this topic (0-1)
1.973 **
1.455 ***
1.342 *
1.478 **
1.216
1.215
Internet user
1.596
1.203
1.399 *
1.443
1.144
-
Cell phone user
1.287
0.881
1.023
0.840
0.806
1.905 *
Facebook visits per month (0-90)
1.001
0.998
0.998
0.996
0.996
1.002
LinkedIn visits per month (0-90)
1.005
1.001
1.003
0.989
0.986 *
0.999
Twitter visits per month (0-90)
1.009
0.999
1.005
1.002
1.000
0.997
Instagram visits per month (0-90)
0.992
1.000
1.001
0.997
1.003
1.001
Pinterest visits per month (0-90)
1.001
0.999
0.998
1.004
1.004
0.994
Number of Facebook friends
1.000
1.000
1.000
1.000 *
1.000
1.000
Status update per month (0-90)
1.036
1.016*
1.020 **
0.996
1.001
1.008
‘Like’ per month (0-90)
1.004
1.007*
1.002
1.005
1.003
1.006 *
Comment per month (0-90)
0.994
0.997
0.997
0.996
0.997
0.994
Sending messages per month (0-90)
1.002
1.004
1.002
1.005
1.004
1.006
Demographics
Married or living with a partner
Black/African-American
Government Surveillance Topic
Media Use
Internet Activities
Facebook Activities
R-squared (Nagelkerke)
0.125 *** 0.081 *** 0.109 *** 0.074 *** 0.067*** 0.108 ***
Notes: N is smaller than 1801 (total sample size) because some respondents did not answer questions about their
demographics or media use; the analysis of co-workers is limited to participants who reported having a full or parttime job; the analysis of Facebook is limited to participants who use Facebook.
*p<.05 **p<.01 ***p<.001
www.pewresearch.org
31
PEW RESEARCH CENTER
Survey questions
August Tracking 2013 / Facebook Survey
Final Topline
9/18/2013
Data for August 7-September 16, 2013
Princeton Survey Research Associates International for
the Pew Research Center’s Internet & American Life Project
Sample: n=1,801 national adults, age 18 and older, including 900 cell phone interviews
Interviewing dates: 08.07.2013—09.16.2013
Margin
Margin
Margin
Margin
Margin
Margin
of
of
of
of
of
of
error
error
error
error
error
error
is
is
is
is
is
is
plus
plus
plus
plus
plus
plus
or
or
or
or
or
or
minus
minus
minus
minus
minus
minus
2.6
2.9
2.7
3.3
3.5
7.2
percentage
percentage
percentage
percentage
percentage
percentage
points
points
points
points
points
points
for
for
for
for
for
for
results
results
results
results
results
results
based
based
based
based
based
based
on
on
on
on
on
on
Total [n=1,801]
all internet users [n=1,445]
all cell phone owners [n=1,636]
all SNS or Twitter users [n=1,076]
Facebook users [n=960]
Twitter users [n=223]
EMINUSE Do
INTMOB
you use the internet or email, at least occasionally?
Do you access the internet on a cell phone, tablet or other mobile handheld device, at
least occasionally?17
Current
May 2013
December 2012i
November 2012ii
September 2012
August 2012iii
April 2012
February 2012
USES INTERNET
DOES NOT USE
INTERNET
80
85
81
85
81
85
82
80
20
15
19
15
19
15
18
20
USES INTERNET
DOES NOT USE
INTERNET
The definition of an internet user varies from survey to survey. Prior to January 2005, internet users were
defined as those who said yes to “Do you ever go online to access the Internet or World Wide Web or to send
and receive email?” From January 2005 thru February 2012, an internet user is someone said yes to either
“Do you use the internet, at least occasionally?” (INTUSE) OR “Do you send or receive email, at least
occasionally?” (EMLOCC). From April 2012 thru December 2012, an internet user is someone said yes to
any of three questions: INTUSE, EMLOCC or “Do you access the internet on a cell phone, tablet or other
mobile handheld device, at least occasionally?” (INTMOB). In May 2013, half the sample was asked
INTUSE/EMLOCC/INTMOB and half was asked EMINUSE/INTMOB. Those May 2013 trend results are for
both forms combined.
17
www.pewresearch.org
32
PEW RESEARCH CENTER
December 2011
August 2011
May 2011
January 2011iv
December 2010v
November 2010vi
September 2010
May 2010
January 2010vii
December 2009viii
September 2009
April 2009
December 2008
November 2008ix
August 2008x
July 2008xi
May 2008xii
April 2008xiii
January 2008xiv
December 2007xv
September 2007xvi
February 2007xvii
December 2006xviii
November 2006xix
August 2006xx
April 2006xxi
February 2006xxii
December 2005xxiii
September 2005xxiv
June 2005xxv
February 2005xxvi
January 2005xxvii
Nov 23-30, 2004xxviii
November 2004xxix
July 2004xxx
June 2004xxxi
March 2004xxxii
February 2004xxxiii
November 2003xxxiv
August 2003xxxv
June 2003xxxvi
May 2003xxxvii
March 3-11, 2003xxxviii
February 2003xxxix
December 2002xl
November 2002xli
October 2002xlii
82
78
78
79
77
74
74
79
75
74
77
79
74
74
75
77
73
73
70
75
73
71
70
68
70
73
73
66
72
68
67
66
59
61
67
63
69
63
64
63
62
63
62
64
57
61
59
www.pewresearch.org
18
22
22
21
23
26
26
21
25
26
23
21
26
26
25
23
27
27
30
25
27
29
30
32
30
27
27
34
28
32
33
34
41
39
33
37
31
37
36
37
38
37
38
36
43
39
41
33
PEW RESEARCH CENTER
September 2002xliii
July 2002xliv
March/May 2002xlv
January 2002xlvi
December 2001xlvii
November 2001xlviii
October 2001xlix
September 2001l
August 2001li
February 2001lii
December 2000liii
November 2000liv
October 2000lv
September 2000lvi
August 2000lvii
June 2000lviii
May 2000lix
61
59
58
61
58
58
56
55
59
53
59
53
52
50
49
47
48
39
41
42
39
42
42
44
45
41
47
41
47
48
50
51
53
52
WEB1-A Next...
Please tell me if you ever use the internet to do any of the following things. Do you
ever use the internet to...[INSERT ITEM; RANDOMIZE; ALWAYS ASK ABOUT FACEBOOK
LAST]?18
Based on all internet users [N=1,445]
TOTAL HAVE
EVER DONE
THIS
---------DID
YEST ERDAY
HAVE NOT
DONE T HIS
DON’T KNOW
REFUSED
18
18
16
16
15
12
13
10
12
8
58
46
37
n/a
n/a
n/a
n/a
8
5
4
n/a
n/a
2
n/a
n/a
5
82
82
84
84
85
88
87
90
88
92
42
54
63
*
*
*
*
*
*
*
*
*
0
*
*
*
0
*
*
0
0
0
0
*
0
*
*
---
Use Twitter
Current
May 2013
December 2012
August 2012
February 2012
August 2011
May 2011
January 2011
December 2010
November 2010
Current
July 2008
August 2006
Prior to January 2005, question wording was “Please tell me if you ever do any of the following when you
go online. Do you ever...[ITEM]?” Unless otherwise noted, trends are based on all internet users for that
survey.
18
www.pewresearch.org
34
PEW RESEARCH CENTER
Use Instagram
Current
December 2012
August 2-5, 2012lx
17
13
12
n/a
n/a
n/a
82
87
88
*
*
1
0
0
0
21
15
12
n/a
n/a
n/a
77
83
87
2
2
1
*
0
*
22
n/a
77
1
*
71
67
n/a
n/a
29
33
0
0
0
*
Use Pinterest
Current
December 2012
August 2-5, 2012
Use LinkedIn
Current
Use Facebook19
Current
lxi
December 13-16, 2012
Q5
Recently, a government program with the aim of collecting information about people’s
telephone calls, emails and other online communications has been in the news. How
interested are you, if at all, in this topic? [READ]
%
Q6
Current
26
34
19
20
1
*
Very interested
Somewhat interested
Not too interested
Not interested at all
(VOL.) Don’t know
(VOL.) Refused
Overall, how KNOWLEDGEABLE would you say you are about the debate surrounding
these government programs aimed at collecting information about people’s calls, emails
and other online communications? Would you say you are... [READ]
%
Current
12
42
28
17
Very knowledgeable
Somewhat knowledgeable
Not too knowledgeable
Not knowledgeable at all
*
*
(VOL.) Don’t know
(VOL.) Refused
December 13-16, 2012 trend was asked of all internet users as a standalone question: "Do you ever use
Facebook?"
19
www.pewresearch.org
35
PEW RESEARCH CENTER
Q7
How much information, if any, have you gotten about this debate from the following
sources? (First,/Next,) how about from [INSERT ITEMS IN ORDER]?
[READ AS NECESSARY: Have you gotten a lot, some, a little, or no information about this
debate from (ITEM)?]
A LOT
SOME
A LITTLE
NONE AT
ALL
DON’T
KNOW
REFUSED
6
14
19
60
1
*
b. TV and radio
31
27
19
22
*
*
c. Friends and family
Item D: Based on Facebook users [N=960]
9
22
25
42
1
*
d. Facebook
Item E: Based on Twitter users [N=223]
10
16
19
54
*
*
e. Twitter
Item F: Based on all internet users [N=1,445]
9
13
13
65
0
0
22
21
12
44
*
*
a.
f.
Q8
Your local print newspaper
Online news sources other than Facebook
or Twitter
Thinking about the debate over the U.S. government's surveillance programs... Do you
FAVOR or OPPOSE a government program to collect nearly all communications in the U.S.
as part of anti-terrorism efforts?
[IF FAVOR/OPPOSE, PROBE:] Do you strongly (favor/oppose) or only somewhat
(favor/oppose) these programs?
%
Current
13
24
22
30
7
3
Strongly favor
Somewhat favor
Somewhat oppose
Strongly oppose
Don’t know
Refused
www.pewresearch.org
36
PEW RESEARCH CENTER
Q9
If the topic of the government’s surveillance programs came up [INSERT FIRST ITEM;
RANDOMIZE], would you be very willing, somewhat willing, somewhat unwilling, or very
unwilling to join in the conversation?
What if this topic came up...[INSERT NEXT ITEM]? [READ AS NECESSARY: Would you be
very willing, somewhat willing, somewhat unwilling, or very unwilling to join in the
conversation?]
Q10
VERY
WILLING
SOMEWHAT
WILLING
SOMEWHAT
UNWILLING
VERY
UNWILLING
DON’T
KNOW
REFUSED
a. At a community meeting
Item B: Based on those employed full or
part-time [N=1015]
26
39
16
16
2
1
b. At work
26
38
15
18
1
1
c.
At a restaurant with friends
32
38
14
14
1
1
d. At a family dinner
Item E: Based on Facebook users
[N=960]
39
34
12
12
1
1
e. On Facebook
Item F: Based on Twitter users [N=223]
15
26
23
34
1
*
f.
14
26
18
38
1
2
On Twitter
Still thinking about the current debate about the government’s surveillance programs... To
what extent do you think [INSERT ITEMS IN ORDER] agree with your views about this
issue? Do you think they mostly agree, somewhat agree, somewhat disagree, or mostly
disagree with your views?
How about [INSERT NEXT ITEM]? [READ AS NECESSARY: Do you think they mostly agree,
somewhat agree, somewhat disagree, or mostly disagree with your views?]
MOSTLY
AGREE
SOMEWHAT
AGREE
SOMEWHAT
DISAGREE
Item A: Based on those who are
married or living with a partner
[N=1,017]
www.pewresearch.org
MOSTLY
DISAGREE
(VOL.)
DOESN'T
APPLY
DON’T KNOW
REFUSED
37
PEW RESEARCH CENTER
Your spouse or partner
53
32
7
4
*
3
1
b. Other family members
35
34
11
6
*
13
2
c. Your close friends
Item D: Based on those
employed full or part-time
[N=1015]
36
36
9
4
1
12
2
d. Your coworkers
23
38
11
5
18
18
2
e. Your neighbors
Item F: Based on Facebook
users [N=960]
17
27
9
6
2
35
3
22
38
10
5
1
20
3
18
32
10
8
9
22
1
a.
f.
The people in your
network on Facebook
Item G: Based on Twitter users
[N=223]
g.
The people who follow you
on Twitter
www.pewresearch.org
38
PEW RESEARCH CENTER
Methods
This report is based on the findings of a Pew Research Center survey on Americans' use of the
Internet. The results in this report are based on data from telephone interviews conducted by
Princeton Survey Research Associates International from August 7 to September 16, 2013, among
a sample of 1,801 adults, age 18 and older. Telephone interviews were conducted in English and
Spanish by landline (901) and cell phone (900, including 482 without a landline phone). For
results based on the total sample, one can say with 95% confidence that the error attributable to
sampling is plus or minus 2.6 percentage points. For results based on Internet users20 (n=1,445),
the margin of sampling error is plus or minus 2.9 percentage points, and for those on Facebook or
Twitter (n=1,076), plus or minus 3.3 points. In addition to sampling error, question wording and
practical difficulties in conducting telephone surveys may introduce some error or bias into the
findings of opinion polls.
A combination of landline and cellular random digit dial (RDD) samples was used to represent all
adults in the United States who have access to either a landline or cellular telephone. Both samples
were provided by Survey Sampling International, LLC (SSI) according to PSRAI specifications.
Numbers for the landline sample were drawn with equal probabilities from active blocks (area
code + exchange + two-digit block number) that contained three or more residential directory
listings. The cellular sample was not list-assisted, but was drawn through a systematic sampling
from dedicated wireless 100-blocks and shared service 100-blocks with no directory-listed landline
numbers.
New sample was released daily and was kept in the field for at least seven days. The sample was
released in replicates, which are representative subsamples of the larger population. This ensures
that complete call procedures were followed for the entire sample. At least 7 attempts were made
to complete an interview at a sampled telephone number. The calls were staggered over times of
day and days of the week to maximize the chances of making contact with a potential respondent.
Each number received at least one daytime call in an attempt to find someone available. For the
landline sample, interviewers asked to speak with the youngest adult male or female currently at
home based on a random rotation. If no male/female was available, interviewers asked to speak
with the youngest adult of the other gender. For the cellular sample, interviews were conducted
with the person who answered the phone. Interviewers verified that the person was an adult and in
a safe place before administering the survey. Cellular sample respondents were offered a post-paid
cash incentive for their participation. All interviews completed on any given day were considered
to be the final sample for that day.
Weighting is generally used in survey analysis to compensate for sample designs and patterns of
non-response that might bias results. A two-stage weighting procedure was used to weight this
dual-frame sample. The first-stage corrected for different probabilities of selection associated with
Internet user definition includes those who use the internet or email at least occasionally or access the
internet on a mobile handheld device at least occasionally.
20
www.pewresearch.org
39
PEW RESEARCH CENTER
the number of adults in each household and each respondent’s telephone usage patterns.21 This
weighting also adjusts for the overlapping landline and cell sample frames and the relative sizes of
each frame and each sample.
The second stage of weighting balances sample demographics to population parameters. The
sample is balanced to match national population parameters for sex, age, education, race,
Hispanic origin, region (U.S. Census definitions), population density, and telephone usage. The
Hispanic origin was split out based on nativity; U.S born and non-U.S. born. The White, nonHispanic subgroup was also balanced on age, education and region. The basic weighting
parameters came from the US Census Bureau’s 2011 American Community Survey data.22 The
population density parameter was derived from Census 2010 data. The telephone usage parameter
came from an analysis of the July-December 2012 National Health Interview Survey.23
Following is the full disposition of all sampled telephone numbers:
Sample Disposition
Landline
40,985
21
Cell
27,000 Total Numbers Dialed
1,669
1,458
15
24,589
1,994
11,260
27.5%
346
94
---10,375
427
15,758
58.4%
Non-residential
Computer/Fax
Cell phone
Other not working
Additional projected not
working numbers
Working
Working Rate
665
3,332
27
7,236
64.3%
142
5,501
16
10,099
64.1%
No Answer / Busy
Voice Mail
Other Non-Contact
Contacted numbers
Contact Rate
328
5,898
1,010
14.0%
1,793
6,776
1,530
15.2%
Callback
Refusal
Cooperating numbers
Cooperation Rate
i.e., whether respondents have only a landline telephone, only a cell phone, or both kinds of telephone.
ACS analysis was based on all adults excluding those living in institutional group quarters (GCs).
SJ, Luke JV. Wireless substitution: Early release of estimates from the National Health
Interview Survey, July-December, 2012. National Center for Health Statistics. June 2013.
22
23Blumberg
www.pewresearch.org
40
PEW RESEARCH CENTER
53
---957
94.8%
67
540
923
60.3%
Language Barrier
Child's cell phone
Eligible numbers
Eligibility Rate
56
901
94.1%
22 Break-off
901 Completes
97.6% Completion Rate
8.4%
9.5% Response Rate
The disposition reports all of the sampled telephone numbers ever dialed from the original
telephone number samples. The response rate estimates the fraction of all eligible respondents in
the sample that were ultimately interviewed. At PSRAI it is calculated by taking the product of
three component rates:
Contact rate—the proportion of working numbers where a request for interview was made
Cooperation rate—the proportion of contacted numbers where a consent for interview was at least
initially obtained, versus those refused
Completion rate—the proportion of initially cooperating and eligible interviews that were
completed
Thus the response rate for the landline sample was 8 percent. The response rate for the cellular
sample was 10 percent.
www.pewresearch.org
41
PEW RESEARCH CENTER
REFERENCES
Boczkowski, P. and E. Mitchelstein (2013). The news gap : when the information preferences of
the media and the public diverge. Cambridge, MIT Press.
Das, S. and A. Kramer (2013). "Self-censorship on Facebook." Proc. of ICWSM 2013: 120-127.
Goel, S., W. Mason, et al. (2010). "Real and Perceived Attitude Agreement in Social Networks."
Journal of Personality and Social Psychology 99(4): 611-621.
Jacobs, L. R., F. L. Cook, et al. (2009). Talking Together: Public Deliberation and Political
Participation in America. Chicago, University of Chicago Press.
Marwick, A. E. and d. boyd (2010). "I Tweet Honestly, I Tweet Passionately: Twitter Users,
Context Collapse, and the Imagined Audience." New Media & Society 13(1): 114-133.
Mitchell, A., J. Holcomb, et al. (2013). News Use across Social Media Platforms. Washingtown,
Pew Research Center.
Noelle-Neumann, E. (1974). "The Spiral of Silence A Theory of Public Opinion." Journal of
Communication 24(2): 43-51.
Oshagan, H. (1996). "Reference Group Influince on Public Expression." International Journal of
Public Opinion Research 8(4): 335-354.
Sunstein, C. R. (2001). Republic.com. Princeton, N.J., Princeton University Press.
Wyatt, R. O., E. Katz, et al. (2000). "Bridging the Spheres: Political and Personal Conversation in
Public and Private Spaces." Journal of Communication 50(1): 71-92.
December 2012 trends based on the 2012 Post-Election Tracking Survey, conducted November 14–
December 9, 2012 [N=2,261, including 908 cell phone interviews].
ii November 2012 trends based on the Gates Library Services Survey, conducted October 15 – November 10,
2012 among those age 16 or older [N=2,252, including 1,125 cell phone interviews].
iii August 2012 trends based on the “Civic Engagement Tracking Survey” conducted July 16–August 7, 2012
[N=2,253, including 900 cell phone interviews].
iv January 2011 trends based on the Pew Internet Project/Project for Excellence in Journalism/Knight
Foundation “Local News survey,” conducted January 12-25, 2011 [N=2,251, including 750 cell phone
interviews].
v December 2010 trends based on the Social Side of the Internet survey, conducted November 23–December
21, 2010 [N=2,303, including 748 cell phone interviews].
vi November 2010 trends based on the Post-Election Tracking Survey 2010, conducted November 3-24, 2010
[N=2,257, including 755 cell phone interviews].
i
www.pewresearch.org
42
PEW RESEARCH CENTER
January 2010 trends based on the Online News survey, conducted December 28, 2009 – January 19, 2010
[N=2,259, including 562 cell phone interviews].
viii December 2009 trends based on the Fall Tracking “E-Government” survey, conducted November 30 –
December 27, 2009 [N=2,258, including 565 cell phone interviews].
ix November 2008 trends based on the Post-Election 2008 Tracking survey, conducted November 20December 4, 2008 [N=2,254].
x August 2008 trends based on the August Tracking 2008 survey, conducted August 12-31, 2008 [N=2,251].
xi July 2008 trends based on the Personal Networks and Community survey, conducted July 9-August 10,
2008 [N=2,512, including 505 cell phone interviews]
xii May 2008 trends based on the Spring Tracking 2008 survey, conducted April 8-May 11, 2008 [N=2,251].
xiii April 2008 trends based on the Networked Workers survey, conducted March 27-April 14, 2008. Most
questions were asked only of full- or part-time workers [N=1,000], but trend results shown here reflect the
total sample [N=2,134].
xiv January 2008 trends based on the Networked Families survey, conducted December 13, 2007-January 13,
2008 [N=2,252].
xv December 2007 trends based on the Annual Gadgets survey, conducted October 24-December 2, 2007
[N=2,054, including 500 cell phone interviews].
xvi September 2007 trends based on the Consumer Choice survey, conducted August 3-September 5, 2007
[N=2,400, oversample of 129 cell phone users].
xvii February 2007 trends based on daily tracking survey conducted February 15-March 7, 2007 [N=2,200].
xviii December 2006 trends based on daily tracking survey, conducted November 30 - December 30, 2006
[N=2,373].
xix November 2006 trends based on Post-Election tracking survey, conducted Nov. 8-Dec. 4, 2006
[N=2,562]. This includes an RDD sample [N=2,362] and a cell phone only sample [N=200]. Results reflect
combined samples, where applicable.
xx August 2006 trends based on daily tracking survey, conducted August 1-31, 2006 [N=2,928].
xxi April 2006 trends based on the Annual Gadgets survey, conducted Feb. 15-Apr. 6, 2006 [N=4,001].
xxii February 2006 trends based on the Exploratorium Survey, conducted Jan. 9-Feb. 6, 2006 [N=2,000].
xxiii December 2005 trends based on daily tracking survey conducted Nov. 29-Dec. 31, 2005 [N=3,011].
xxiv September 2005 trends based on daily tracking survey conducted Sept. 14-Oct.13, 2005 [N=2,251].
xxv June 2005 trends based on the Spyware Survey, conducted May 4-June 7, 2005 [N=2,001].
xxvi February 2005 trends based on daily tracking survey conducted Feb. 21-March 21, 2005 [N=2,201].
xxvii January 2005 trends based on daily tracking survey conducted Jan. 13-Feb.9, 2005 [N=2,201].
xxviii November 23-30, 2004 trends based on the November 2004 Activity Tracking Survey, conducted
November 23-30, 2004 [N=914].
xxix November 2004 trends based on the November Post-Election Tracking Survey, conducted Nov 4-Nov 22,
2004 [N=2,200].
xxx July 2004 trends based on the “Selective Exposure” survey, conducted June 14-July 3, 2004 [N=1,510].
xxxi June 2004 trends based on daily tracking survey conducted May 14-June 17, 2004 [N=2,200].
xxxii March 2004 trends based on “Weak Ties” survey conducted February 17-March 17, 2004 [N=2,200].
xxxiii February 2004 trends based on daily tracking survey conducted February 3-March 1, 2004 [N=2,204].
xxxiv November 2003 trends based on daily tracking survey conducted November 18-December 14, 2003
[N=2,013].
xxxv August 2003 trends based on ‘E-Government’ survey conducted June 25-August 3, 2003 [N=2,925].
xxxvi June 2003 trends based on ‘Internet Spam’ survey conducted June 10-24, 2003 [N=2,200].
xxxvii May 2003 trends based on daily tracking survey conducted April 29-May 20, 2003 [N=1,632].
xxxviii March 3-11, 2003 trends based on daily tracking survey conducted March 3-11, 2003 [N=743].
xxxix February 2003 trends based on daily tracking survey conducted February 12-March 2, 2003 [N=1,611].
xl December 2002 trends based on daily tracking survey conducted Nov. 25–Dec. 22, 2002 [N=2,038].
xli November 2002 trends based on daily tracking survey conducted October 28-November 24, 2002
[N=2,745].
xlii October 2002 trends based on daily tracking survey conducted October 7-27, 2002 [N=1,677].
vii
www.pewresearch.org
43
PEW RESEARCH CENTER
September 2002 trends based on daily tracking survey conducted September 9-October 6, 2002
[N=2,092].
xliv July 2002 trends based on ‘Sept. 11th-The Impact Online’ survey conducted June 26-July 26, 2002
[N=2,501].
xlv March/May 2002 trends based on daily tracking surveys conducted March 1-31, 2002 and May 2-19,
2002.
xlvi January 2002 trends based on a daily tracking survey conducted January 3-31, 2002.
xlvii December 2001 trends represent a total tracking period of December 1-23, 2001 [N=3,214]. This tracking
period based on daily tracking surveys conducted December 17-23, 2001 and November 19-December 16, 2001.
xlviii November 2001 trends represent a total tracking period of November 1-30, 2001 [N=2,119]. This
tracking period based on daily tracking surveys conducted October 19 – November 18, 2001 and November
19 – December 16, 2001.
xlix October 2001 trends represent a total tracking period of October 1-31, 2001 [N=1,924]. This tracking
period based on daily tracking surveys conducted September 20 – October 1, 2001, October 2-7, 2001,
October 8-18, 2001, and October 19 – November 18, 2001.
l September 2001 trends represent a total tracking period of September 1-30, 2001 [N=742]. This tracking
period based on daily tracking surveys conducted August 13-September 10, 2001, September 12-19, 2001
and September 20 – October 1, 2001.
li August 2001 trends represent a total tracking period of August 13-31, 2001 [N=1,505]. This tracking period based on a
daily tracking survey conducted August 13-September 10, 2001.
lii February 2001 trends based on a daily tracking survey conducted February 1, 2001-March 1, 2001
[N=2,096].
liii December 2000 trends based on a daily tracking survey conducted December 2-22, 2000 [N=2,383].
liv November 2000 trends based on a daily tracking survey conducted November 2 – December 1, 2000 [N=6,321].
lv October 2000 trends based on a daily tracking survey conducted October 2 – November 1, 2000 [N=3,336].
lvi September 2000 trends based on a daily tracking survey conducted September 15 – October 1, 2000 [N=1,302].
lvii August 2000 trends based on a daily tracking survey conducted July 24 – August 20, 2000 [N=2,109].
lviii June 2000 trends based on a daily tracking survey conducted May 2 – June 30, 2000 [N=4,606].
lix May 2000 trends based on a daily tracking survey conducted March 1 – May 1, 2000 [N=6,036].
lx August 2-5, 2012 trends based on an omnibus survey conducted August 2-5, 2012 [N=1,005, including 405
cell phone interviews]. Omnibus survey not conducted as a tracking survey.
lxi December 13-16, 2012 trends based on an omnibus survey conducted December 13-16, 2012 [N=1,006,
including 405 cell phone interviews]. Omnibus survey not conducted as a tracking survey.
xliii
www.pewresearch.org